Impacts of congestion pricing and reward strategies on automobile travelers' morning commute mode shift decisions
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Impacts of congestion pricing and reward strategies on automobile travelers’ morning
commute mode shift decisions
Yaping Lia, Yuntao Guob, Jian Lua*, Srinivas Peetac,d
a
School of Transportation, Southeast University, 2 Southeast University Road，Jiangning District，,
Nanjing 211189, China
b
Lyles School of Civil Engineering/NEXTRANS Center, Purdue University, 3000 Kent Avenue ， West
Lafayette, IN 47906, USA
c
School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA
d
H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta,
GA 30318, USA
*
Corresponding author
ABSTRACT
This paper studies the impacts of congestion pricing and reward strategies on automobile
travelers’ morning commute mode shift decisions using stated preference travel mode choice
data of over 1,000 automobile travelers collected in Beijing inner districts. To address the
complex impacts of these strategies on automobile travelers, a stage-based model framework
is developed to analyze their mode shift decision-making process (whether they will shift from
using automobile to public transit, biking or walking or continue using automobile) under these
strategies. Four multilevel structural equation models are created for participants using
automobile (personal vehicle and/or taxi) as the most common mode of transportation
(hereafter referred to as “more habitual automobile travelers”) and those using automobile as
second most common mode (hereafter referred to as “less habitual automobile travelers”) under
each strategy. Model estimation results show that the impacts of latent psychological factors
on mode shift decisions under congestion pricing and reward strategies are significantly
different between more and less habitual automobile travelers. The results also show that
congestion pricing strategies are more effective than reward strategies in promoting mode shifts
among more habitual automobile travelers, while the opposite is observed for less habitual
automobile travelers. This study provides insights for designing congestion pricing and reward
strategies and illustrates the importance of developing complementary modules that target
numerous factors in different stages of the mode shift decision-making process to effectively
promote mode shifts from automobile to sustainable travel modes in China.
1. Introduction
With rapid economic growth, vehicle ownership has increased drastically in China over
the past decade. From 2012 to 2016, private vehicle stock has increased from 72.2 to 146.4
million and this growth trend is expected to continue in the future (Motor Vehicle Pollution
Prevention Report of China, 2016). Growing vehicle ownership has created many challenges
in metropolitan regions such as growing traffic congestion, environmental pollution and energy
consumption which have negative impacts on human health and environment (Zhang and
Crooks, 2012; Tommy and Friman, 2015; Guo et al., 2018). There is a critical need to develop
effective strategies to promote mode shifts from automobile to sustainable travel modes such
as public transit, bike and walk (hereafter referred to as “mode shifts to sustainability”).
Various pricing strategies, such congestion pricing and reward strategies, have been
considered to promote such mode shifts in China. Megacities such as Beijing, Shanghai,
Shenzhen and Guangzhou have evaluated the feasibility of implementing congestion pricing
strategies to promote mode shifts by increasing automobile travel cost (Sun et al., 2016).However, these strategies were not implemented in these cities after evaluation because of
various social concerns such as equity and public acceptance (Link, 2015). As an alternative to
congestion pricing strategies, reward strategies, are considered as innovative and effective
pricing strategies to promote mode shifts to sustainability by providing monetary incentives to
travelers for using sustainable travel modes (Khademi, 2016). Empirical evidence has shown
that both strategies can potentially promote mode shifts to sustainability (Harbeck, et al., 2017;
Ettema and Verhorf, 2006). However, none of the previous studies has evaluated the potential
for implementing reward strategies in China, analyzed their impacts on mode shift decisions,
compared these impacts with those of congestion pricing strategies on mode shift decisions or
understood the potential similarities and dissimilarities of these impacts on travel mode shift
decisions of more and less habitual automobile travelers.
Apart from using congestion pricing and reward strategies to promote mode shifts to
sustainability, previous studies also show that providing personalized, trip- and mode-specific
information such as pollution emission information (e.g., amount of CO2 emissions) and
physical activity information (e.g., calories burnt) associated with each mode can also
encourage such mode shifts (Jariyasunant, et al., 2015; Börjesson, et al., 2016; Guo and Peeta,
2017; Guo et al., 2017; Sunio and Schmöcker, 2017). For example, “Quantified Traveler” is a
Web-based app that provides feedback in terms of cost, calories, time and emissions based on
a given origin-destination trip to promote mode shifts to sustainability (Jariyasunant, et al.,
2015). These results illustrate that integrating personalized, trip- and mode-specified
information with congestion pricing and reward strategies can potentially increase their
effectiveness.
This study aims at evaluating the effectiveness of congestion pricing and reward strategies
on morning commute mode shift decisions (whether they will shift from using automobile to
public transit, biking or walking or continue using automobile) of more and less habitual
automobile travelers in China. A stage-based modeling framework is developed to capture the
impacts of psychological factors on travelers’ mode choice behavioral changes. To achieve
these objectives, a web-based stated preference survey for Beijing inner districts in China is
designed. Participants were asked to select their travel mode choice before and after the
implementation of congestion pricing and reward strategies for their morning commute. Eight
modes of transportation were considered, including personal vehicle, taxi (including traditional
taxi and ridesharing such as DiDi taxi1), bus transit, subway transit, electric bike, shared bike
(including docked bike and dockless bike such as Mobike2), personal bike and walk. Both
shared and personal bikes are manually operated. Personal vehicle and taxi are considered as
automobile modes and the rest as sustainable travel modes. An interactive mode choice
information map is designed for the survey to provide participants with travel-related
information, including travel time, travel cost/reward, amount of CO2 emissions, amount of
particulate matter emissions, walk steps and calories burnt associated with using each mode for
a given origin and destination, once they have entered home and work addresses. By providing
these types of information, we can also ensure that all participants can have similar levels of
familiarity with the modes available. Four multilevel structural equation models using the
1
DiDi taxi is one type of taxi hailing service via a smartphone application in China. This taxi-hailing app (like Uber) uses
GPS technology to allow users to locate taxicabs that are nearby on their handheld device, and then users can send the
required information to book a taxi.
2
Mobike is one type of bike sharing service via a smartphone application in China. This bike-sharing app allows users to
pick up and leave a bike at their convenience.stage-based modeling framework are estimated based on both more and less habitual
automobile travelers’ stated preference mode shift decisions under each strategy while capture
the potential correlations among these responses. This study provides a comprehensive and in-
depth understanding of the impacts of congestion pricing and reward strategies on morning
commute mode shift decisions. These results and insights can assist decision-makers to develop
intervention strategies to complement pricing strategies that target automobile travelers’
various stages of mode shift decision-making process to increase public acceptance and
improve the effectiveness of various pricing strategies to promote mode shifts to sustainability
in China.
The remainder of this paper is organized as follows. The next section outlines the relevant
theoretical background and proposes a conceptual model for mode shift decisions under pricing
strategies. After that, the survey design and descriptive statistics are analyzed. The model
estimation results are then presented and discussed. The paper concludes with some comments
and insights.
2. Literature review
The theoretical background of the decision-making process can be categorized into three
main clusters: (1) the theory of planned behavior (TPB), which characterizes the decision-
making process as a process of forming the intention to adopt certain behavior (Ajzen, 1985),
(2) norm activation model (NAM), which explains the decision-making process as a
motivational basis for the realization of altruistic behavior (Hunecke et al., 2001), and (3) self-
regulation model, which describes the decision-making process as a transition through a
sequence of volitional phases (Schwarzer, 2008; Fu et al, 2012).
The TPB and NAM have been widely used to model the travel mode shift decision-making
process in the literature (Bamberg et al, 2007; De Groot et al, 2008; Klöckner and Matthies,
2009; Hsiao and Yang, 2010; Eriksson and Forward, 2011; Mann and Abraham, 2012;
Onwezen et al, 2013; Nordfjærn et al, 2014; Lois, Moriano, & Rondinella, 2015; Guo and Peeta,
2015). A key assumption in these studies is that a person’s choices are governed by his/her
behavioral intention or pro-social motives. However, empirical evidence (e.g. Armitage and
Conner, 2001; Bamberg and Möser, 2007; Susan, et.al, 2009) has shown that there is a gap
between intention and behavior (referred to as “intention-behavior gap”). It implies that
behavioral changes such as mode shifts to sustainability requires travelers to not only form
strong behavioral intentions or pro-social motives but also develop skills and strategies to
control the temptation of reverting back to old behavior (such as using automobile) and break
down barriers to implement a set of behaviors (such as using sustainable travel modes).
To bridge such gaps, some recent studies (e.g., Bekkum and Elizabeth, 2011; Fu, et.al.
2012) introduced conceptual self-regulation models based on the assumption that
implementation intention is the foundation of the volitional phase which can bridge the
intention-behavior gap. Bamberg (2013a, 2013b) proposed a more cumulative theoretical
framework which integrates the stage concept with TPB and NAM. This model describes the
behavioral change process as a series of four stages: (a) pre-decisional stage, (b) pre-actional
stage, (c) actional stage and (d) post-actional stage. Goal intention, behavioral intention and
implement intention are three transition points between stages. The process can be described
as follows:
• In the pre-decisional stage, an individual becomes aware of the problems and re-
evaluates his/her habitual behavior. This can create the goal intention (i.e. the intention to act)that transitions to the pre-actional stage. Such intention can be captured by personal norm (i.e.
personal value system) which is correlated with a traveler’s awareness of consequences (i.e.,
the consequences of shifting or not shifting to sustainable travel modes), sense of responsibility
to making such shift, and social norms (i.e. perceived social pressure to engage/not engage in
shifting from automobile to sustainable travel modes).
• Once the goal intention is established, the individual transitions to the pre-actional stage
and starts to select a new behavioral alternative and prepare for change. This leads to behavioral
intention (i.e. self-commitment to change to a new alternative). Attitude (i.e.
favorable/unfavorable view of using each travel mode.) and perceived behavioral control (i.e.
perceived ability to shift to sustainable travel modes) are the two main social-cognitive factors
that promote the formation of behavioral intention.
• In the actional stage, the individual initiates and implements necessary actions to make
the behavioral change. The enactment of such change is facilitated by implementation intention.
Concrete planning abilities and the ability to remove psychological barriers (action/coping
planning) are essential factors to implementation intention. Then, the decision-making process
transitions to the post-actional stage.
• In the post-actional stage, the individual will evaluate what he/she has achieved and
decide whether to maintain the new behavior. The individual’s confidence can influence his/her
ability to maintain the new behavior and resume the new behavior after a relapse (so-called
self-efficacy). This is critical to maintaining the new behavior.
The aforementioned behavioral change model provides a framework to quantify the
impacts of psychological factors on the behavioral change by breaking from habitual travel
behavior and/or forming new ones. However, to the best of authors’ knowledge, none of the
existing studies have applied this model to understand the impacts of monetary
incentives/disincentives (i.e., rewarding mode shift to sustainability or penalizing the usage of
automobile through pricing strategies) on travel mode shift decision-making process in China.
In this study, a stage-based mode shift decision-making model under pricing strategies is
proposed, as shown in Fig. 1. The model structure is adopted from Bamberg (2013a) with two
major differences highlighted in green. First, the post-actional stage is substituted by an
adaptive stage without further predictors. This is because pricing strategies have yet to be
implemented in Beijing and only stated preferences are available. Second, in the pre-actional
stage, perception of pricing strategies (i.e. individual’s cognitive capacity to evaluate the
strategies) has been added as an additional predictor. As shown in previous studies (Sun et al,
2016), a traveler’s willingness to adopt an alternative travel mode under pricing strategies is
associated with his/her perception of these strategies (such as perceived effectiveness, freedom,
fairness and acceptability).Pre-decisional Stage Pre-actional Stage Actional Stage Adaptive Stage
Awareness of Ascribed Attitude Action planning
consequence responsibility
Personal Goal Perceived Behavioral Implementation Mode shift
norm intention behavioral control intention intention decision
Perception of
Social norm Coping planning
pricing strategy
Determine whether
Re-evaluate current behavior Select new behavior alternative Make a decision
implement it
Fig. 1. Proposed conceptual model.3. Methods
To understand travelers’ mode shift decision-making process and evaluate the proposed
model, an experiment is designed using stated preference survey method. In the survey,
participants who live and/or work in Beijing inner districts (Fig. 2) were asked to answer a wide
range of questions (Fig. 3) related to each component of the proposed conceptual model.
Participants were expected to choose between “continue to travel by car/taxi” and “switch to
sustainable travel mode” under three congestion pricing strategies and three reward strategies.
The following subsections include more details related to the description of the study area
(section 3.1), survey design (section 3.2), how the survey was implemented and some of the
descriptive statistics (section 3.3), and structural equation modeling method (section 3.4).
3.1 Study area
As shown in Fig. 2, the Beijing inner districts, which include Xicheng, Dongcheng,
Haidian, Chaoyang, Shunyi, and Shijingshan, were selected as the study area. In these districts,
travelers often experience high congestion during morning peak hours due to increasing
automobile ownership. According to the Annual Report on Traffic Development in Beijing
(2016), the commute mode shares of automobile, bus and subway are around 31.9%, 25% and
25%, respectively. The average vehicle speed during the morning peak hour in the study area is
about 14.7 km/h (or 9.1 mph), and the travel time during peak hour is twice as much as the
travel time under free-flow conditions. There is an urgent need to reduce automobile usage and
promote mode shifts to sustainability. Participants are recruited based on the inclusion criteria
that they must live or work in Beijing inner districts, and the most or second common mode of
morning commute (go to work/school) is automobile (include car or taxi).
(a) Beijing inner districts (b) Residential population of each district (in millions)
Fig. 2. Survey area.
3.2 Survey design
The structure of the questionnaire is presented in Fig. 3. It consists of three main parts:
participants’ sociodemographic characteristics, current morning commute mode choice and
morning commute mode choice preference under six pricing strategies and psychological
factors related to the morning commute mode shift decision-making process. Six pricing
strategies include three congestion pricing strategies and three reward strategies. In the first part,
questions related to personal and household information, and household mobility resources
were asked.Personal and household Household mobility
information resources
- Gender
Socio-demographic information - Age - Number of e-bikes
- Education level - Number of bikes
• Personal and household information - Household size - Number of cars
• Household mobility resources - Personal income - Availability of public
- Employment status transport cards
Worker or - Work hour flexibility
student
Sustainable
Most common commute modes Second common
mode commute mode
Commute mode choice preference
car or taxi car or taxi
• Mode choice decision under pricing
strategy scenarios Commuting mode choice
decision
- Congestion pricing scenarios
- Reward strategy scenarios
Pre-decisional stage Pre-actional stage
- Awareness of consequence
- Attitude
Psychological characteristics - Ascribed responsibility
- Perceived behavioral control
- Social norm
• Pre-decisional stage - Perception of pricing strategies
- Personal norm
• Pre-actional stage - Behavioral intention
- Goal intention
• Actional stage
Actional stage
- Action planning
- Coping planning
- Implementation intention
Fig. 3. Structure of questionnaire.
The second part aims to capture participants’ current morning commute behavior and
stated mode shift decisions under six pricing strategies. The eight available modes are car, taxi,
bus transit, subway transit, electric bike, personal bike, shared bike and walk. Participants were
asked to select their most and second most common travel modes for morning commute. Only
those who use automobile (i.e., car/taxi) as their most or second most common commute mode
are included in the study. Based on their answers to these two questions, participants are divided
into two groups: more and less habitual automobile travelers. By doing so, the potential
similarities and dissimilarities of stated mode choice preference under various pricing strategies
between travelers with different automobile usage habit strengths can be analyzed. These results
will facilitate the designs of complementary strategies to pricing strategies for different sub-
populations to promote mode shifts to sustainability. To understand their stated preference of
mode choice under various pricing strategies, an interactive online mode choice information
tool is designed, as shown in Appendix A. After participants enter their residential location and
work/school location on the map, it provides mode choice options and information related to
travel distance, travel time, travel cost, amount of CO2 emissions, amount of particulate matter
emissions, number of steps walked, the amount of activity calories burned and possible
rewards/penalties for each mode under each proposed pricing strategy. The estimation of
pollution emission information (amount of CO2 emissions and amount of particulate matter
emissions) and physical activity information (number of steps walked and the amount of
activity calories burned) are based on previous studies (Table 1). The estimated amount of CO2emissions includes both exhaust emissions and indirect emissions of fuel consumed (Ma et al,
2011; Grazi et al, 2008) and the particulate matter emission factor is calculated based on an
empirical study on emission factors of vehicle exhausts in Beijing (Fan et al, 2015). The number
of steps walked is based on the stride length and average height of Beijing residents (Kanchan
et al, 2015), and calories burnt are estimated based on a diet calculator (Golan, Y., 1998).
Table 1 Data used to calculate pollution emission information and physical activity
information
Electric Personal Shared
Car Taxi Bus Subway Walk
bike bike bike
CO2 emission factor (g/km) 178.6 178.6 73.8 9.1 69.6 0 0 0
Particulate matter emission
1 1 0.4 0.05 0.3 0 0 0
factor (mg/km)
Walk steps (steps/km) 0 0 0 0 0 0 0 2000
Activity calories (Kcal/km) 0 0 0 0 0 27 27 38
Participants were asked to select their preferred morning commute travel mode under three
congestion pricing strategies (penalized with 5-yuan, 15-yuan and 25-yuan for using
automobile). Similarly, under the three reward strategies, participants were asked to select their
preferred travel mode if they are rewarded with 1-yuan, 1.5-yuan and 2-yuan for using
sustainable travel modes. The amount of monetary award given to travelers is designed to cover
part or all of the travel costs if the participants want to use bus or subway. If a traveler has the
bus pass (it only costs a 20-yuan deposit to obtain, and most travelers have it), the ticket costs
1-yuan. The starting price for subway ticket is 3-yuan, and a 50% discount is received if
monthly cost of subway tickets is over 150-yuan. The estimated subway ticket cost per traveler
per trip is 4.3-yuan. The reward amount and congestion cost are not created equal because
congestion pricing strategies require a large amount of initial investment, but can generate
revenue for government to cover the cost in the long run, while reward strategies can only
increase government spending. It is not financially feasible for the government to set lower
congestion price or higher reward values.
The third part of the survey is designed to capture psychological factors related to mode
shifts. The details of these questions are presented in Table 2. Most of these questions are
developed based on Bamberg (2013a), including participants’ awareness of consequence (2
questions), ascribed responsibility (1 question), social norm (2 questions), personal norm (2
questions), goal intention (1 questions), attitude (3 questions) and perceived behavioral control
(2 questions), using 5-point Likert scales between strongly disagree and strongly agree.
Additional ones adapted from the literature (Sun et al., 2016; Bamberg, 2013b) were also asked
related to participants’ perception of pricing strategies (4 questions), behavioral intention (1
question) and implementation intention (1 question) under each pricing strategy. In addition, to
capture action planning and coping planning, the method developed by Hsieh et.al (2017) is
used. For action planning, participants were asked about when, where and how to act if they
shift from automobile to sustainable travel modes, which includes how to search schedule and
how to check required travel time information. For coping planning, the ability to overcome
specific barriers and perception of different potential barriers are included. The ratio of the
perceived level of barriers to the perceived ease of overcoming barriers is used to weigh coping
planning.Table 2 Survey questions for psychological variables
Category Latent variable Item Observed variable
Pre- Awareness of AC1 a Traffic congestion and pollution will become more serious with increasing automobile usage.
decisional consequence AC2 a There is no benefit to personal health if we use automobile to travel more.
stage Ascribed responsibility ARa I feel personally responsible for the problems related to automobile usage.
Social norm SN1 a People who are important to me think that it is good to commute using sustainable travel modes.
SN2 a Most people who are important to me expect me to reduce automobile usage.
Personal norm PN1 a I feel personally obliged to reduce automobile usage as much as possible.
PN2 a Regardless of what other people do, I have a moral obligation to reduce automobile usage.
Goal intention GI a I intend to think over how to reduce automobile usage in the future.
Pre- Perceived behavioral PBC1 b How much control do you have over whether you commute by automobile or not?
actional control PBC2 a For me, it is easy to use sustainable travel modes more frequently to commute.
stage Attitude ATT1 a I like to travel by sustainable travel modes.
ATT2 a Using sustainable travel modes more make me feel good.
ATT3 a If I reduce automobile usage, I will have positive influence on alleviating the problems caused by automobile
usage.
Perception of pricing PP1-C c Do you perceive that each of the three the proposed congestion pricing strategies would alleviate traffic
strategy congestion and pollution?
PP1-R c Do you perceive that each of the three proposed reward strategies would alleviate traffic congestion and
pollution?
PP2-C d Do you perceive that each of the three proposed congestion pricing strategies would be fair to you?
PP2-R d Do you perceive that each of the three proposed reward strategies would be fair to you?
PP3-C e Do you perceive that each of the three proposed congestion pricing strategies would affect your freedom to
choose travel modes?
PP3-R e Do you perceive that each of the three proposed reward strategies would affect your freedom to choose any travel
mode?
PP4-C f Do you perceive that each of the three proposed congestion pricing strategies would be acceptable?
PP4-R f Do you perceive that each of the three proposed reward strategies would be acceptable?Behavioral Intention BI1-C a I intend to use more sustainable travel modes frequently for everyday trips under each of the three proposed
congestion pricing strategies.
BI1-R a I intend to use more sustainable travel modes frequently for everyday trips under each of the three proposed
reward strategies.
a
Actional Action planning AP1 I know how to search the schedule if I change to public transit.
stage AP2 a I know how to check the travel time if I change to sustainable travel modes.
g, h
Coping planning CP1 Could inflexibility of departure time be a barrier to shift from automobile to public transit? f and, Is it easy for
you to overcome it? g
g, h
CP2 Could large travel time be a barrier to shift from automobile to sustainable travel modes? f and, Is it easy for you
to overcome it? g
CP3 g, h Could difficulty of reaching places not near a transit station be a barrier to shift from automobile to public transit?
f
and, Is it easy for you to overcome it? g
CP4 g, h Could lack of freedom to travel be a barrier to shift from automobile to sustainable travel modes? f and, Is it easy
for you to overcome it? g
CP5 g, h Could the possibility of harsh weather be a barrier to shift from automobile to sustainable travel modes? f and, Is
it easy for you to overcome it? g
CP6 g, h Could inconvenience of carrying luggage be a barrier to shift from automobile to sustainable travel modes? f and,
Is it easy for you to overcome it? g
a
Implementation II-C I will travel by sustainable travel modes on my next morning commute under each of the three proposed
intention congestion pricing strategies.
a
II-R I will travel by sustainable travel modes on my next morning commute under each of the three proposed reward
strategies.
a b
Scales 1-5: from “strongly disagree” (1) to “strongly agree” (5). Scales 1-5: from “not at all” (1) to “complete control” (5).
c d
Scales 1-5: from “not at all” (1) to “very effective” (5). Scales 1-5: from “very unfair” (1) to “very fair” (5).
e f
Scales 1-5: from “not at all” (1) to “very freedom” (5). Scales 1-5: from “not at all” (1) to “very acceptable” (5).
g h
Scales 1-5: from “no, not at all” (1) to “yes, largely” (5) Scales 1-5: from “not at all” (1) to “very easy” (5).3.3 Survey implemented and descriptive statistics
The questionnaire is distributed online by Sojump Survey Company
(http://www.sojump.com) which possesses more than 2.6 million sample resources with diverse
sociodemographic characteristics. Potential participants who meet the selection criteria in the
respondent pool maintained by Sojump received an email invitation to join the study. Out of
the 2500 questionnaire distributed, 1135 completed questionnaires were returned between mid-
June and mid-July of 2017.
Table 3 Sociodemographic characteristics of the target population and participants
Characteristic Target All samples More habitual Less habitual p-value2
Population1 (N=1135) automobile automobile
travelers (N=557) travelers (N=578)
Gender
Male 50.9% 58.6% 64.8% 52.6%
0.000
Female 49.1% 41.4% 35.2% 47.4%
Age
18-24 12.5% 19.3% 18.7% 19.9%
25-34 26.2% 45.7% 46.5% 45.0%
35-44 18.4% 27.0% 26.2% 27.7% 0.000
45-54 16.4% 7.1% 7.2% 7.1%
>55 26.5% 0.8% 1.4% 0.3%
Education level
High school
43.3% 7.5% 5.9% 9.0%
diploma or lower
College degree 46.9% 74.3% 74.7% 74.0% 0.021
Post-graduate
degree or above 9.8% 18.2% 19.4% 17.0%
Personal monthly income (Yuan)
10000 26.0% 26.0% 32.9% 19.4%
Household size
1 22.7% 12.6% 11.8% 13.3%
2 30.7% 16.6% 14.2% 19.0%
0.008
3 29.0% 41.0% 42.0% 40.1%
≥4 17.6% 29.7% 32.0% 27.5%
Employment status
Public sector
22.0% 28.3% 25.5% 31.0%
employee
Private sector
65.3% 56.7% 58.7% 54.7%
employee 0.001
Self-
8.1% 6.7% 6.8% 6.7%
employment
Students - 8.3% 9.0% 7.6%
1
Data from Beijing Census Bureau 2015.
2
Chi-square test result on demographic characteristics between more and less habitual automobile travelers.The sociodemographic characteristics of the population in the target area, survey samples
(N = 1135), and more (N = 557, 49.1%) and less (N = 578, 50.9%) habitual automobile travelers
are shown in Table 3. Compared to the sociodemographic characteristics of the target
population, our sample comprises a larger percentage of men, aged between 25 and 44, with
college degree or above, with relatively high monthly income (the median monthly wage in
Beijing Urban area was about 7086 yuan per month in 2015 based on the National Bureau of
Statistics (2016)), or a household with more than 3 family members. In addition, more habitual
automobile travelers are a larger percentage in the aforementioned categories compared to less
habitual automobile travelers. These differences exist because all respondents use automobile
as the first or second most common mode choice; so, they often belong to middle class or above
households with a higher income and larger family size, especially for the more habitual
automobile travelers.
Continue to travel by car/taxi Switch to sustainable travel modes
100%
90% 86.5%
79.0% 78.8%
80% 73.5%
70% 64.2%
60% 55.3%
50% 44.7%
40% 35.8%
30% 26.5%
21.0% 21.2%
20% 13.5%
10%
0%
5-yuan 15-yuan 25-yuan 1-yuan 1.5-yuan 2-yuan
Congestion pricing strategies Reward strategies
(a) More habitual automobile travelers
Continue to travel by car/taxi Switch to susutinable travel modes
100%
90% 85.3% 87.0% 88.2%
82.5% 83.3%
80%
70%
61.2%
60%
50%
38.8%
40%
30%
20% 17.5% 14.7% 16.7%
13.0% 11.8%
10%
0%
5-yuan 15-yuan 25-yuan 1-yuan 1.5-yuan 2-yuan
Congestion pricing strategies Reward strategies
(b) Less habitual automobile travelers
Fig.4. Mode shift decisions under congestion pricing and reward strategies.
Fig. 4 illustrates participants’ stated mode shift decisions under six congestion pricing and
reward strategies of more and less habitual automobile travelers. For participants who choose
to shift to sustainable travel modes under congestion pricing and reward strategies, about 35.5%,
18.9% and 17.5% of them want to shift to subway, bus and biking, respectively. The resultsshow that when the congestion fee is 5 yuan, over half of the participants want to shift from
automobile to sustainable transportation modes, while in developed countries such as in Sweden
(Karlström & P. Franklin, 2009) or Netherlands (Tillema et al, 2013), the congestion fee needs
to be much higher to promote mode shifts. However, our results are consistent with one recent
study (Linn et al, 2016) in Beijing which estimates that 95% automobile travelers will be gave
up driving (or taking a taxi) if congestion fee is 8 Yuan. There are three main reasons for the
differences observed between Beijing and cities in some developed countries. First, Beijing has
relatively complete intra-city public transportation service, which consists of subway, bus, and
bicycle sharing systems. For example, over 20 percent of participants who choose to shift to
subway from automobile will experience a shorter travel time, and only about 20 percent of
them will experience a 25 percent or higher increase in travel time. Second, rising concerns
related to health risks and environmental pollution caused by intensifying city smog also
contribute to a higher willingness to shift modes. Over 67 percent of the participants in our
study have a strong perception that health risks and environmental pollution will become more
serious with increased automobile usage. This is consistent with the literature (Sun et al, 2017)
which shows that more than 63 percent of Beijing residents have strong awareness of smog and
environmental pollution. Third, the estimated average subway and bus fares per traveler per
morning commute trip are 4.3-yuan and 1.5-yuan, respectively, which are much lower than
under automobile usage and the rewards provided are sufficient to cover most or all of the
subway and bus fares.
Four additional observations can be made from Fig. 4. First, consistent with the literature,
most participants are more likely to shift to sustainable travel modes as the penalty/reward
increases. Second, the percentage difference between travelers making mode shifts under 15-
yuan and 25-yuan congestion fees is not very significant, as also under 1.5-yuan and 2-yuan.
One reason is that participants with higher income are more likely to be less sensitive to the
amount of penalty/reward provided, but are more likely to value more other benefits associated
with using automobile such as the sense of freedom and shorter travel times. Another reason is
that some travelers with relatively longer morning commute times (sometimes over one hour
using automobile) are captive automobile travelers and the commute times under sustainable
travel modes are too long (sometimes double or triple that amount) for them to be viewed as
credible alternatives. Third, a smaller percentage of more habitual automobile travelers is more
likely to shift to sustainable travel modes under congestion pricing strategies compared to that
under reward strategies, while the opposite is observed for less habitual automobile travelers.
A possible reason is people who are more habitual automobile travelers are less sensitive to
rewards associated with shifting to sustainable travel modes, while being more sensitive to
penalties for using automobile. This is different from the findings of some past studies (e.g.
Tillema et al, 2013) in which automobile travelers are more willing to reduce car usage under
a reward strategy than that under a congestion pricing strategy. This is likely because the reward
provided is relatively lower compared to other perceived benefits of using automobile. Fourth,
congestion pricing and reward strategies are more effective in promoting less habitual
automobile travelers to shift to sustainable travel modes, especially reward strategies. This is
because less habitual automobile travelers are using sustainable travel modes as their most
common morning commute mode choice and it is easier for them to shift to modes that they are
more familiar with. Considering that in developing countries such as China, travelers have onlyjust started becoming habitual automobile travelers unlike most developed countries with high
automobile ownership, it may be more beneficial to implement pricing strategies in these
countries to impede the habit of using automobile and promote the usage of sustainable travel
modes when the automobile habit is not very strong.
3.4 Multilevel Structural equation modeling (ML-SEM)
As each participant responded to each of the three congestion pricing strategies and reward
strategies, it is important to factor the potential correlation among these three responses.
Multilevel structural equation modeling framework Rabe-Hesketh et al, 2004a; Rabe-Hesketh
et al, 2004b; Rabe-Hesketh et al, 2007) is used in this study to capture such potential
correlations. A multilevel structural equation model is a statistical technique to model the
relationships of multiple independent and dependent variables varying at different levels.
Within this framework, multilevel structural equation models combine both measurement
models and structural models. The measurement models are used to study the possible
interrelationships among a set of observed variables (response to survey questions) of one latent
variable, which can be written as follows:
L Ml
v = b x + å å hm( )lm( )¢ zm( )
l l l
(1)
l = 2 m =1
where v represents the log odds for observed indicators, x is explanatory variable
associated with fixed effect b . The mth latent variable at level l, hm(l ) , is multiplied by a linear
combination of explanatory variable lm(l )¢ zm(l ) . Note that the variables of a multivariate response
are treated as level 1 units in a multilevel dataset (the original units become level 2 units).
The structural model is used to capture relationships between different latent variables,
which can be represented as:
η = Bη + ζ (2)
where B is a regression parameter matrix for the relations among the latent variables η ,
and ζ is a vector of errors. Details of ML-SEM theory, model identification issues, estimation
procedures, and model evaluation can be found in Rabe-Hesketh et al (2004b).
In this study, the survey data is modeled as a two-level structural equation model where
individual’s choices of pricing strategies scenarios are considered as a lower level (level 1)
which are nested in individual at level 2, as shown in Fig.5.
Individual i Individual i Level 2
5-yuan 15-yuan 25-yuan 1-yuan 1.5-yuan 2-yuan Level 1
（a）Congestion pricing strategies （b）Reward strategies
Fig.5. Survey data structure
Four separate ML-SEM models for more and less habitual automobile travelers’ mode
shift decision-making process under congestion pricing and reward strategies are estimatedusing STATA. The latent variables used in level 1 includes perception of pricing strategies,
behavioral intention, implementation intention and mode choice decision. The mode choice
decision is a binary choice between “continue to travel by car/taxi” and “switch to sustainable
travel mode” for each strategy. The model optimization processes utilize modification indices
to improve the model fit. Model fitness is evaluated using χ2/df, The Tucker-Lewis index (TLI),
Comparative Fit Index (CFI), and Root Means Square Error of Approximation (RMSEA)
(Schreiber et al, 2006). Based on the model fitness of each estimation iteration, modification
indices are used to adjust the originally hypothesized model until the goodness-of-fit indices
indicate a reasonable fit.
4. Results and discussion
4.1 Reliability analysis
Before conducting SEM analysis, we first examined the reliability of the latent variables
expected to be used in SEM. Cronbach's α is calculated to assess scale reliabilities and the
internal consistency of different questions within a latent variable in the conceptual model. The
means and Cronbach's α of all latent variables are shown in Table 4. Note that the mode shift
decision in the proposed conceptual model is whether a participant will continue using
automobile or shift to sustainable travel modes. The mode shift decision is modeled as a binary
variable, with choosing automobile as 1 and choosing sustainable travel modes as 0. All
Cronbach's α levels of the variables, which have more than one observed variable, are greater
than the acceptable level (α>0.70) in practice (Nunnally, 1978), indicating a high level of
reliability and internal consistency. Moreover, the composite reliability of all measures in each
model is also greater than the acceptable level (i.e. 0.70). Overall, these results show that the
measurement questions in this study possess adequate reliability.
4.2 ML-SEM analysis of mode shift decision-making process under pricing strategies
The estimation results of the measurement models are shown in Table 5. Standardized
coefficients of measurement model paths illustrate the relationship between observed variables
and latent variables and only statistically significant paths (pTable 4(a) Mean and reliability analysis of latent variables (For more habitual automobile travelers)
Congestion pricing strategy Reward strategy
5-yuan 15-yuan 25-yuan 1-yuan 1.5-yuan 2-yuan
Latent variables Mean Mean Mean Mean Mean Mean
α α α α α α
(SD) (SD) (SD) (SD) (SD) (SD)
Awareness of AC1 3.70 (1.15) 3.76 (1.15) 3.76 (1.15) 3.76 (1.15) 3.76 (1.15) 3.76 (1.15)
0.79 0.79 0.79 0.79 0.79 0.79
consequence AC2 3.76 (1.20) 3.70 (1.20) 3.70 (1.20) 3.70 (1.20) 3.70 (1.20) 3.70 (1.20)
Ascribed responsibility AR 3.42 (1.27) - 3.42 (1.27) - 3.42 (1.27) - 3.42 (1.27) - 3.42 (1.27) - 3.42 (1.27) -
SN1 3.42 (1.20) 3.42 (1.20) 3.42 (1.20) 3.42 (1.20) 3.42 (1.20) 3.42 (1.20)
Social norm 0.81 0.81 0.81 0.81 0.81 0.81
SN2 3.16 (1.22) 3.16 (1.22) 3.16 (1.22) 3.16 (1.22) 3.16 (1.22) 3.16 (1.22)
PN1 3.50 (1.16) 3.50 (1.16) 3.50 (1.16) 0.79 3.50 (1.16) 3.50 (1.16) 3.50 (1.16)
Personal norm 0.79 0.79 0.79 0.79 0.79
PN2 3.28 (1.06) 3.28 (1.06) 3.28 (1.06) 3.28 (1.06) 3.28 (1.06) 3.28 (1.06)
Goal intention GI 3.56 (1.15) - 3.56 (1.15) - 3.56 (1.15) - 3.56 (1.15) - 3.56 (1.15) - 3.56 (1.15) -
ATT1 3.03 (1.17) 3.03 (1.17) 3.03 (1.17) 3.53 (1.10) 3.53 (1.10) 3.53 (1.10)
Attitude ATT2 3.67 (1.05) 0.72 3.67 (1.05) 0.72 3.67 (1.05) 0.72 2.20 (1.84) 0.72 2.20 (1.84) 0.72 2.20 (1.84) 0.72
ATT3 3.09 (1.50) 3.09 (1.50) 3.09 (1.50) 3.65 (1.05) 3.65 (1.05) 3.65 (1.05)
Perceived behavioral PBC1 3.66 (1.05) 3.66 (1.05) 3.66 (1.05) 3.66 (1.05) 3.66 (1.05) 3.66 (1.05)
0.73 0.73 0.73 0.73 0.73 0.73
control PBC2 3.41 (1.14) 3.41 (1.14) 3.41 (1.14) 3.41 (1.14) 3.41 (1.14) 3.41 (1.14)
PP1 3.31 (1.43) 3.31 (1.43) 3.31 (1.43) 3.01 (1.23) 3.01 (1.23) 3.01 (1.23)
Perception of pricing PP2 2.92 (1.25) 2.92 (1.25) 2.92 (1.25) 3.20 (1.15) 3.20 (1.15) 3.20 (1.15)
0.73 0.73 0.73 0.79 0.79 0.79
strategy PP3 2.96 (1.28) 2.96 (1.28) 2.96 (1.28) 3.42 (1.06) 3.42 (1.06) 3.42 (1.06)
PP4 2.95 (1.30) 2.95 (1.30) 2.95 (1.30) 3.41 (1.20) 3.41 (1.20) 3.41 (1.20)
Behavioral intention BI 3.12 (1.10) - 3.44 (1.10) - 3.54 (1.07) - 2.96 (0.68) - 3.11 (0.86) - 3.25 (1.37) -
AP1 3.54 (1.09) 3.54 (1.09) 3.54 (1.09) 3.54 (1.09) 3.54 (1.09) 3.54 (1.09)
Action Planning 0.91 0.91 0.91 0.91 0.91 0.91
AP2 3.56 (1.10) 3.56 (1.10) 3.56 (1.10) 3.56 (1.10) 3.56 (1.10) 3.56 (1.10)
CP1 1.88 (0.72) 1.88 (0.72) 1.88 (0.72) 1.88 (0.72) 1.88 (0.72) 1.88 (0.72)
CP2 2.40 (1.55) 2.40 (1.55) 2.40 (1.55) 2.40 (1.55) 2.40 (1.55) 2.40 (1.55)
CP3 1.84 (1.34) 1.84 (1.34) 1.84 (1.34) 1.84 (1.34) 1.84 (1.34) 1.84 (1.34)
Coping Planning 0.76 0.76 0.76 0.76 0.76 0.76
CP4 1.84 (0.75) 1.84 (0.75) 1.84 (0.75) 1.84 (0.75) 1.84 (0.75) 1.84 (0.75)
CP5 1.59 (1.05) 1.59 (1.05) 1.59 (1.05) 1.59 (1.05) 1.59 (1.05) 1.59 (1.05)
CP6 1.47 (0.86) 1.88 (0.72) 1.88 (0.72) 1.88 (0.72) 1.88 (0.72) 1.88 (0.72)
Implementation intention II 3.05 (1.34) - 3.23 (1.08) - 3.43 (1.53) - 2.98 (0.69) - 3.12 (1.06) - 3.34 (1.25) -
Automobile decision AD 0.36 (0.65) - 0.21 (0.44) - 0.14 (0.45) - 0.45 (0.48) - 0.27 (0.45) - 0.21 (0.56) -
Sustainable travel modes
STMD 0.64 (0.54) - 0.79 (0.44) - 0.86 (0.42) - 0.55 (0.36) - 0.74 (0.25) - 0.79 (0.42) -
decision
Composite reliability 0.86 0.87 0.87 0.88 0.88 0.89Table 4(b) Mean and reliability analysis of latent variables (For less habitual automobile travelers)
Congestion pricing strategy Reward strategy
5-yuan 15-yuan 25-yuan 1-yuan 1.5-yuan 2-yuan
Latent variables Mean Mean Mean Mean Mean Mean
α α α α α α
(SD) (SD) (SD) (SD) (SD) (SD)
Awareness of AC1 3.83 (1.00) 3.83 (1.00) 3.83 (1.00) 3.83 (1.00) 3.83 (1.00) 3.83 (1.00)
0.74 0.74 0.74 0.74 0.74 0.74
consequence AC2 3.89 (1.02) 3.89 (1.02) 3.89 (1.02) 3.89 (1.02) 3.89 (1.02) 3.89 (1.02)
Ascribed responsibility AR 3.59 (1.15) - 3.59 (1.15) - 3.59 (1.15) - 3.59 (1.15) - 3.59 (1.15) - 3.59 (1.15) -
SN1 3.70 (1.02) 3.70 (1.02) 3.70 (1.02) 3.70 (1.02) 3.70 (1.02) 3.70 (1.02)
Social norm 0.79 0.79 0.79 0.79 0.79 0.79
SN2 3.45 (1.04) 3.45 (1.04) 3.45 (1.04) 3.45 (1.04) 3.45 (1.04) 3.45 (1.04)
PN1 3.93 (0.89) 3.93 (0.89) 3.93 (0.89) 0.78 3.93 (0.89) 3.93 (0.89) 3.93 (0.89)
Personal norm 0.78 0.78 0.78 0.78 0.78
PN2 3.68 (1.01) 3.68 (1.01) 3.68 (1.01) 3.68 (1.01) 3.68 (1.01) 3.68 (1.01)
Goal intention GI 3.91 (0.87) - 3.91 (0.87) - 3.91 (0.87) - 3.91 (0.87) - 3.91 (0.87) 3.91 (0.87) -
ATT1 3.21 (1.12) 3.21 (1.12) 3.21 (1.12) 3.32 (1.11) 3.32 (1.11) 3.32 (1.11)
Attitude ATT2 3.67 (1.01) 0.78 3.67 (1.01) 0.78 3.67 (1.01) 0.78 3.49 (0.87) 0.78 3.49 (0.87) 0.78 3.49 (0.87) 0.78
ATT3 3.93 (0.89) 3.93 (0.89) 3.93 (0.89) 2.75 (1.12) 2.75 (1.12) 2.75 (1.12)
Perceived behavioral PBC1 3.95 (0.86) 3.95 (0.86) 3.95 (0.86) 3.95 (0.86) 3.95 (0.86) 3.95 (0.86)
0.76 0.76 0.76 0.76 0.76 0.76
control PBC2 3.92 (0.88) 3.92 (0.88) 3.92 (0.88) 3.92 (0.88) 3.92 (0.88) 3.92 (0.88)
PP1 3.48 (1.34) 3.48 (1.34) 3.48 (1.34) 3.17 (1.11) 3.17 (1.11) 3.17 (1.11)
Perception of pricing PP2 3.12 (1.12) 3.12 (1.12) 3.12 (1.12) 3.38 (1.01) 3.38 (1.01) 3.38 (1.01)
0.73 0.73 0.73 0.78 0.78 0.78
strategy PP3 3.14 (1.11) 3.14 (1.11) 3.14 (1.11) 3.59 (1.12) 3.59 (1.12) 3.59 (1.12)
PP4 3.15 (1.18) 3.15 (1.18) 3.15 (1.18) 3.60 (1.00) 3.60 (1.00) 3.60 (1.00)
Behavioral intention BI 3.46 (1.13) - 3.56 (0.95) - 3.89(0.43) - 3.82 (0.83) - 3.90 (1.04) - 3.97 (1.21) -
AP1 3.85 (0.91) 3.85 (0.91) 3.85 (0.91) 3.85 (0.91) 3.85 (0.91) 3.85 (0.91)
Action Planning 0.88 0.88 0.88 0.88 0.88 0.88
AP2 3.87 (0.85) 3.87 (0.85) 3.87 (0.85) 3.87 (0.85) 3.87 (0.85) 3.87 (0.85)
CP1 1.61 (0.99) 1.61 (0.99) 1.61 (0.99) 1.61 (0.99) 1.61 (0.99) 1.61 (0.99)
CP2 1.72 (1.06) 1.72 (1.06) 1.72 (1.06) 1.72 (1.06) 1.72 (1.06) 1.72 (1.06)
CP3 1.58 (0.84) 1.58 (0.84) 1.58 (0.84) 1.58 (0.84) 1.58 (0.84) 1.58 (0.84)
Coping Planning 0.82 0.82 0.82 0.82 0.82 0.82
CP4 1.59 (1.02) 1.59 (1.02) 1.59 (1.02) 1.59 (1.02) 1.59 (1.02) 1.59 (1.02)
CP5 1.53 (0.92) 1.53 (0.92) 1.53 (0.92) 1.53 (0.92) 1.53 (0.92) 1.53 (0.92)
CP6 1.53 (0.89) 1.53 (0.89) 1.53 (0.89) 1.53 (0.89) 1.53 (0.89) 1.53 (0.89)
Implementation intention II 3.29 (1.06) - 3.46 (1.03) - 3.52 (0.79) - 3.51 (1.11) - 3.62 (0.98) - 3.76 (0.75) -
Automobile decision AD 0.39 (1.15) - 0.18 (0.88) - 0.15 (0.62) - 0.17 (1.04) - 0.13 (0.77) - 0.12 (0.69) -
Sustainable travel modes
STMD 0.61 (0.94) - 0.72 (0.79) - 0.85 (0.75) - 0.83 (1.21) - 0.87 (0.86) - 0.88 (0.93) -
decision
Composite reliability 0.87 0.87 0.87 0.88 0.88 0.89Table 5 Measurement model estimate for testing construct of latent variables
Measurement model path Standardized coefficient
More habitual automobile Less habitual automobile
travelers travelers
Congestion Reward Congestion Reward
pricing strategy pricing strategy
strategy strategy
Awareness of consequence→AC1 0.73† 0.73† 0.67† 0.67†
Awareness of consequence→AC2 0.87*** 0.87*** 0.90*** 0.90***
Ascribed responsibility→AS 1 1 1 1
Social norm→SN1 0.66† 0.66† 0.67† 0.67†
Social norm→SN2 0.70** 0.70** 0.71** 0.71**
Personal norm→PN1 0.63† 0.63† 0.65† 0.65†
Personal norm→PN2 0.73*** 0.73*** 0.76*** 0.76***
Goal intention→GI 1 1 1 1
† † †
Perceived behavioral control→PBC1 0.70 0.70 0.72 0.72†
Perceived behavioral control→PBC2 0.81* 0.81* 0.84* 0.84*
Attitude→ATT1 0.51† 0.51† 0.56† 0.56†
Attitude→ATT2 0.47** 0.47** 0.75*** 0.75***
Attitude→ATT3 0.86*** 0.88*** 0.63** 0.63**
† † †
Perception of pricing strategy→PP1 0.41 0.62 0.35 0.62†
Perception of pricing strategy→PP2 0.92*** 0.93*** 0.93*** 0.88***
Perception of pricing strategy→PP3 0.86** 0.82** 0.89** 0.87**
Perception of pricing strategy→PP4 0.79** 0.72** 0.75** 0.75**
Behavioral Intention→BI 1 1 1 1
† † †
Action planning→AP1 0.92 0.92 0.91 0.91†
Action planning→AP2 0.84*** 0.84*** 0.80*** 0.80***
† † †
Coping planning→CP1 0.51 0.51 0.63 0.63†
Coping planning→CP2 0.66*** 0.66*** 0.67*** 0.67***
Coping planning→CP3 0.49* 0.49* 0.54** 0.54**
Coping planning→CP4 0.52** 0.52** 0.62** 0.62**
Coping planning→CP5 0.63** 0.63** 0.76** 0.76**
Coping planning→CP6 0.69** 0.69** 0.65** 0.65**
Implementation intention→II 1 1 1 1
Automobile decision→AD 1 1 1 1
Sustainable travel modes decision→
1 1 1 1
STMD
†
denotes a path for which the unstandardized coefficient is set to 1 as a reference item, and given this
constraint, other paths of remaining questions within a latent variable can be then estimated.
* denotes pPre-decisional Stage Pre-actional Stage Actional Stage Adaptive Stage
Perceived
Awareness of .70**/.70**
Ascribed Attitude behavioral Action
consequence responsibility control planning
Automobile
.20***/.21*** .36***/.37*** decision
.35**/.35** n.s./n.s. .22***/.29***
-.24***/-.21***
Personal Behavioral Implementation
norm
.79***/.79*** Goal intention .58***/.33***
intention .54***/.32***
intention
.24***/.21***
.62***/.62*** .16*/n.s. .28*/.31*
Sustainable travel
modes decision
Perception of Coping
Social norm pricing strategies planning
*pPre-decisional Stage Pre-actional Stage Actional Stage Adaptive Stage
Perceived
Awareness of .53**/.53**
Ascribed Attitude behavioral Action
consequence responsibility control planning
Automobile
.23***/.34*** .34***/.36*** decision
.57***/.57*** n.s./n.s. .25***/.28***
-.25***/-.28***
Personal Behavioral Implementation
.72***/.72*** Goal intention .55***/.59*** .46***/.48***
norm intention intention
.25***/.28***
.55***/.55*** .06*/n.s. .36*/.38*
Sustainable travel
modes decision
Perception of Coping
Social norm pricing strategies planning
*pBy contrast, the awareness of consequence plays a more important role in strengthening such
obligation among less habitual automobile travelers compared to more habitual automobile
travelers. A probable reason is that more habitual automobile travelers are less sensitive to the
negative consequences caused by using automobile because using automobile is more like a
habit to them and the potential consequences of using automobile are not in their decision-
making process. Instead, guilt or other negative emotions induced by external social pressure
can stimulate them to assess their current habit (i.e. using automobile for morning commute)
and promote shift to sustainable travel modes. Compared to social norms and awareness of
consequence, the impacts of ascribed responsibility on personal norm are not significant. A
possible reason is that habitual travelers may believe that they should not be personally
responsible to problems caused by automobile usage as they consider themselves not using
automobile much. They may consider that these problems should be solved by government and
people who use automobile more often than themselves.
In the pre-actional stage, a traveler’s impulse from goal intention is translated into his or
her more concrete behavioral intentions. Attitude and perceived behavior control have a
positive impact on behavioral intention, and these impacts are slightly larger under reward
strategies compared to that of congestion pricing strategies. This suggests that travelers who
have a strong favorable evaluation of sustainable travel modes (i.e. attitude) and/or perceive
shifting to sustainable travel modes is easy, are more likely to have a stronger behavioral
intention to shift to sustainable travel modes under the reward strategy compared to that of
congestion pricing. This is consistent with the results in previous studies (Khademi et al. 2014;
Schall and Mohnen, 2017) that potential monetary gain for using sustainable travel modes can
more effectively reinforce positive affection and attitude towards these modes which leads to
the formation of behavioral intentions to shift to these modes. These results also suggest that
attitude has a larger impact on behavioral intention of less habitual automobile travelers
compared to that of more habitual automobile travelers, while the opposite is true for perceived
behavior control. This suggests that improving the perceived degree of favorable evaluation
may be more effective in promoting shifts to sustainable travel modes for less habitual
automobile travelers compared to that for more habitual automobile travelers under pricing
strategies. For more habitual automobile travelers, the perceived ease of shifting to sustainable
travel modes are more effective in promoting shifts to sustainable travel modes compared to
the perceived degree of favorable evaluation. Apart from attitude and perceived behavior
control, perception of pricing strategies also has a positive impact on behavioral intention under
congestion pricing, which implies that the behavioral intention is stronger under higher
perceived effectiveness, fairness, freedom and acceptance of congestion pricing strategies.
However, perception of pricing strategies under reward strategies are not statistically
significantly correlated with behavioral intention. This suggests although some people have a
strong positive perception of reward strategies, they have a relatively fixed arrival time and
high penalties for late arrival for morning commute may discourage them from forming
behavioral intention to shift to sustainable travel modes.
In the actional stage, behavioral intention and two types of planning ability (action
planning and coping planning) have significant positive impacts on implementation intention.
Action planning and coping planning have a larger direct impact on more habitual automobile
travelers’ implementation intention under the impacts of congestion pricing, while these
impacts are slightly larger under reward strategies for less habitual automobile travelers. Itindicates that the ability to plan and overcome barriers are more effective in promoting less
habitual automobile travelers to form implementation intention to shift to sustainable travel
modes under pricing strategies, especially under reward strategies. Additionally, the impacts of
action planning on implementation intention are significantly larger than the impacts of coping
planning on implementation intention. This implies that knowledge on how to overcome
barriers is more strongly associated with implementation intention to shift to sustainable travel
modes.
Moreover, the implementation intention has a significant negative impact on participants
who continue to make the decision to use automobile, and a significant positive impact on those
who plan to switch to sustainable travel modes. For more habitual automobile travelers, the
impact of implementation intention on automobile usage decision or sustainable travel modes
decision is larger under the congestion pricing strategy than under the reward strategy, while
the opposite holds for less habitual automobile travelers. The relationship between goal
intention, behavioral intention and implementation intention also follows a similar pattern.
These results illustrate that more habitual automobile travelers are more sensitive to penalties
associated with using automobile than the rewards associated with shifting to sustainable travel
modes in their mode shift decision-making process under pricing strategies. For less habitual
automobile travelers, the monetary gain may enhance their willingness to take the initiative to
choose sustainable travel modes under congestion pricing and reward strategies.
The estimation results show that the impacts of latent psychological variables on
promoting mode choice behavioral change intentions are different between more and less
habitual automobile travelers under congestion pricing and reward strategies in different stages
of the mode shift decision-making process. These differences illustrate the need for
policymakers to design differential complementary intervention strategies to stimulate the
intention to switch from automobile to more sustainable modes at different stages of their
decision-making process for more and less habitual automobile travelers to improve the
effectiveness of congestion pricing and reward strategies. For less habitual automobile travelers,
intervention strategies that enhance awareness of consequence, positive attitude towards
shifting to sustainable travel modes, perceived behavioral control, action planning and coping
planning can potentially improve the effectiveness of both congestion pricing and reward
strategies. For more habitual automobile travelers, complementary intervention strategies that
focus on improving their perceived external social pressure can increase the effectiveness of
both strategies. If reward strategies are implemented, complementary intervention strategies
are needed to enhance positive attitudes towards shifting to sustainable travel modes, action
planning and coping planning.
4.3 Identification of psychological determinants of mode shift decisions under pricing
strategies
The direct impacts of different psychological variables revealed from the model structure
were presented in Fig. 5. Further, the total impacts of each psychological factor on mode shift
decision were also analyzed to explore which variables have larger impact on mode choice
behavior under congestion pricing and reward strategies. For this purpose, the indirect impacts
of each variable on the mode shift decision are calculated first. For example, for the mode shift
decision-making process model under congestion pricing for more habitual automobile
travelers, the direct impact of action planning on the sustainable travel modes decision is 0,You can also read
